Registered Data

[02322] Low discrepancy point sets inspired by Sudoku hypercubes

  • Session Time & Room : 3E (Aug.23, 17:40-19:20) @G601
  • Type : Contributed Talk
  • Abstract : Monte Carlo methods are effective to avoid the "Curse of Dimensionality," while not perfect since their convergences are late. To overcome the weakness, quasi-Monte Carlo methods have been developed. Some of the methods use low discrepancy point sets called $(T, M, S)$-nets. In this talk, I present a new construction procedure of $(T, M, S)$-nets from orthogonal arrays as an application of the extension of Sudoku to higher dimensions named Sudoku hypercubes.
  • Classification : 05B15, 65C05
  • Format : Talk at Waseda University
  • Author(s) :
    • Shigetaka Taga (University of Tsukuba)